CHAUMENY Jean Luc, KANTY Piotr, REITMEIER Tobias
1.INTRODUCTION
Soil mixing is a construction method which mixes the existing subsoil with a binder to increase the strength and stiffness of the ground. It can also be used for decreasing the permeability and creating tightening elements. In both cases quality control is an important, not so clear and easy topic.
1.1 Method statement
The production sequence is quite simple. A drilling rig drills a hole while slurry – consisting of water and binder – is pumped through the auger into the ground. The special designed auger mixes the soil with the slurry and creates a homogeneous column with defined characteristics. If needed, the drilling and mixing process is repeated several times. The quantity of injected slurry and number of mixing sequences is defined depending on the base material (ground) and on the needed quality (strength, stiffness, permeability).
1.2 Use of soil mixing
The most important characterization of wet soil mixing is that the main quantity of the product is the existing subsoil which is improved. As a result, there is no excavated material and also less material transport as for concrete or gravel. As long as water is available this method is independent of material supply since only cement which can be stored in silos needs to be delivered on site. Soil mixing is an easy and cost-efficient method for improving the characteristics of the subsoil to create tight and load bearing elements. Therefore, it is used for diaphragm walls for retaining systems but also as deep foundation for buildings and soil strengthening. Soil mixing can also be used as a supporting system for construction pits. The advantage of using the existing soil mainly may also be the biggest disadvantage. Depending on the original ground, the used quantity of cement and many other factors, uniaxial strengths between 1 N/mm² and 7 N/mm² is usually reached. Compared to concrete the characteristics of this material have a certain spread which makes quality control important and not easy as well.
2 QUALITY CONTROL
As already mentioned quality control of soil mixing is important as the final product consists not of homogeneous materials such as concrete used for piles or columns. Before starting a project, preliminary tests are performed. For his purpose material is dug on site and mixed in a laboratory with binder and water. Depending on usage in future the created samples are tested for the uniaxial compressive strength or permeability. Based on these results the production on site starts with a defined cement content and grout density which can be adapted after first samples from the field have been tested. This second step is a most difficult one because reliable results of samples taken from the ground are needed.
2.1 Sampling methods
Basically, there are three common ways for getting samples that can be used for quality control of wet soil mixing elements which have different advantages and disadvantages:
– Wet sampling method: material is taken out of the mixed element as long as it is still wet and soft. It is poured into forms to create cylinder or cubic moulds which are stored under laboratory conditions until they are tested (usually UCS). While with this method it is easy to create a big number of samples one problem is that the samples can be very different in behaviour to the original element. Although they consist of the same material compaction and also curing process can be very different.
– Core samples: samples are drilled out of a hard column on a defined depth. The extracted cores show in-situ condition of soil mixing element in a quite representative way. After the samples have been prepared they can also be tested as wet samples as well. The biggest problem of this method is that it is very difficult to get proper core samples drilled out of elements with an UCS of less than 5 N/mm². Moreover, it is expansive and time consuming (drilling must be in bonded, quite stiff material) which means it is not a suitable method for getting a huge number of samples.
– Liner samples: a pipe is placed in a wet column until the material inside gets hard. Thereafter the pipe is pulled out. After several days the pipe is opened and cylinder samples can be cut off for testing. This method combines the two others mentioned before. As all of these methods have positive and negative points, all of them can be used in the right time, as long as it is kept in mind when the results of the samples are evaluated.
2.2 Regulations and standards
Basic European standard is the EN 14679. This document defines most relevant points for deep mixing projects. Beginning with the method itself, the design, all phases of execution and also the control of deep mixing is mentioned. Although this standard seems to be quite complete and treats all aspects of this construction method, it has the weakness that it does not instruct a certain way of quality control which can be used as standard method and can guarantee a quality standard for all parties. The Austrian guideline (ÖBV) for soil-mixing brings good ideas and also methods for sampling on soil mixing columns. It also contains a way for statistical analyses of how to deal with the spread of samples. It is also mentioning a certain way to consider samples not passing the original design criteria. Nevertheless a mandatory regulation for taking sample is missing.
3.DISCUSSING AND COMPARING UCS-TESTS
During execution of a soil-mixing project a lot of samples have been taken and tested. When receiving the first UCS-test results after a short construction period, it is obvious that there is a certain spread of the values. There are several explanations for that phenomena:
– errors in sample preparation
– cracks of the sample before testing because of not suitable storing
– insufficient testing equipment
– natural heterogeneity of material
As most samples are taken as wet-sample in the above mentioned method the easiest explanation is that the production of samples was wrong. For example, bad compacted or the extracted material has a varying content of cement. In Figure 1 the results of the UCS-test of one jobsite are shown. The blue crosses are 53 wet samples, taken out of the soil-mixing column. The green crosses symbolize the UCS-values of 11 core samples which were drilled out of columns. All results were tested after a period of at least 56 days. Although there is a huge spread between the UCS-test results it is interesting that the average value for both testing methods has the same range between 3 N/mm² and 4 N/mm². A logical explanation for the fact that the spread of the wet samples is bigger are different procedures in production of samples.
Outgoing from this experience results of five other job sites have been compared. In four out of six jobsites the spread between the lowest and the highest UCS-value from wet samples is greater than from the core ones. The results shown in Table 1 were the reason for making several tests on two different soil mixing columns. The results of the UCS-test of the core samples are shown in figure 2. As can be seen, there is a spread of minimum and maximum UCS of 2 or even 3. While in one column there seems to be a cluster at 4,8 N/mm² the other column has no cluster.
4. YOUNG MODULUS ESTIMATION
Columns material stiffness can be checked on the same samples as the strength. The deformations can be checked during uniaxial compression or triaxiale tests. Knowing the stress and strain of the sample Young modulus can be estimated. Below (figure 3) two example of stress-strain curves obtained from uniaxial compression test are shown. Two possible moduli are shown on the graphs E50 (the inclination of curve in 50% of the strength) and Eur (inclination of curve during unloading and reloading). According the recent state of knowledge the values of E50 modulus are correlated to fcm and fck strength [1]. There is a lack of such a correlation for Eur modulus which is in most cases higher than E50 [2], [3]. However due to results of the simulations shown in [3] the use of Eur modulus can give results which are closer to field results (stiffness obtained from load tests). The beginning of the test curve is neglected due to bedding error correction [4].